[PS-2.13] Mind wandering can support automaticity

Noelle, D. C.

Cognitive and information Sciences, University of California, Merced

Our task focus, and associated state of cognitive control, is often difficult to maintain. Distracting experiences and thoughts seem to easily displace our attention from the task at hand (Smallwood & Schooler, 2006). This kind of "mind wandering" might be seen as a frequent failure of cognitive control processes (Killingsworth & Gilbert, 2010), but it is reasonable to consider the possibility that this instability of states of cognitive control is actually adaptive (Schooler, et al., 2011). Indeed, some have argued that too much cognitive control can hinder performance (Beilock, et al., 2002; Cohen & Maunsell, 2011), and there is some evidence that mind wandering supports creative problem solving (Sio & Ormerod, 2009). A largely unexplored potential benefit of nstable cognitive control states involves their role in skill learning and automaticity. Control signals actively maintained in the frontal cortex can be seen as modulating more posterior brain circuits, overcoming the prepotent tendencies of those circuits in a context-appropriate manner (Cohen, Dunbar, & McClelland, 1990). For those posterior circuits to eventually learn to automatically enact a skill, they must overcome their initial dependence on frontal modulation. When neural plasticity largely depends on performance, as in error-correction and reinforcement learning paradigms, these circuits will continue to depend on frontal control signals as long as they assist in
performance. Thus, unstable control representations can provide a means for "removing the training wheels" on a system as it learns to perform a skill without frontal involvement. In this work, we build on a previous model of motor skill learning (Gupta & Noelle, 2007) in order to demonstrate how instability of frontal representations can
lead to automaticity by incrementally reducing dependence on executive control mechanisms.